When execution becomes unhealthy

When execution becomes unhealthy
10 minutes by James Samuel

Teams can slide from healthy execution to unhealthy execution fast. Common signs include too much work in progress, constant priority changes, and deadlines that nobody believes in. The fix is not to push the team harder but to examine the system around them. Leaders should reduce work in progress, control how new work enters the queue, make priorities explicit, and help teams give estimates that actually mean something.

[Webinar] How to stop rewriting your agent’s code
sponsored by Unblocked

You write the rules files, maintain the skills, and paste in the Slack thread your agent missed. But hand-curated context goes stale, and when the sources disagree, your agent can’t tell which is right. Join us live on Oct 28 (FREE) to see how a context engine helps agents get it right the first time.

The religion of speed
6 minutes by Graybeard

Speed has become something people treat as a virtue in itself, mistaking constant motion for real progress. But most "fast" work just skips the thinking, then spends far longer cleaning up the mess. Understanding the problem first, then deciding, then moving is usually faster in the end. Reality is patient, and the bill for skipped thinking always comes due.

Torn down to last
10 minutes by Mike Fisher

PayPal's massive C++ class, PIMP, grew to over 5,000 methods before engineers spent two years untangling it — work that could have been done far earlier. Japan's Ise Grand Shrine offers a better model: rebuild every 20 years so experienced craftsmen can teach the next generation directly. The real risk isn't forgetting knowledge, it's never transferring it in the first place. AI makes this more urgent by generating code nobody fully understands.

The AI bill is the cheap part
5 minutes by Aviv Ben-Yosef

Cutting AI costs gets attention, but it's rarely the real problem draining engineering teams. Surveys show over 15% of engineer time disappears into bad processes, firefighting, misalignment, and misunderstandings. Spend two weeks tracking where time actually goes before optimizing anything. AI only amplifies what your team already does, so fixing how people work together will unlock far more value than trimming token costs.

I am often wrong
2 minutes by Boris Cherny

Good problem solving follows a clear loop: gather information, define the problem, plan a simple solution, then act. New information means revisiting earlier steps, which can feel messy but is actually healthy. The most common failure is not defining the problem or solution clearly enough. Being wrong is useful because it sharpens your thinking and gets you to the right answer faster.

industry

• How Amazon pushes up prices all across the Internet
• Muse Wonder: A frontier model for personal & professional AI agents
• Anthropic will be most ridiculous IPO of year, analyst says
• GTM is getting compressed
• TypeSafe AI raises Series A

security

• Online ads weaponised to deliver zero-click spyware, researchers say
• AI agents aren’t going rogue. Frontier labs are skipping security basics.
• Hackers heat French supermarket fridges, sparking fires and spoiling food
• Why aren't AI companies being charged for hacking?
• 123456 password used in massive Danish CPR data breach

And the most popular article from the last issue was:

• How AI has changed my engineering leadership role